Pool secretly sidelining hashing-power?

[aim6i]

Hi!

I want to make clear that this is not meant to be a accusation of anybody, instead rather some questions that I hope somebody can prove as wrong.  

Would it be possible for a big pool to secretly sideline some of the total hashing power to another pool or to solo mining?
A big pool could easily divert a very small percentage of its hashing power without anybody noticing. This "side"-hashing-power could be connected to another pool or simply to solo mining which would make it nearly undetectable.

As far as I understand this matter it could be theoretically possible as it is the pool deciding which piece of work the miners receive.

I was thinking about this theory lately because I did notice a significant difference between deepbit and BTCGuild over the past 5 weeks. I was pointing my miners to both pools, 50:50. However, with deepbit I almost always had a better BTC/24h ratio than with BTCGuild EVEN THOUGH deepbit subtracts a 3%-fee! In total it made a difference of around 1 btc which is not huge but very annoying nevertheless.  

Of course there is a lot of luck in the game, but in a time span of 5 weeks this variance should be heavily diminished.

[1714793474]

1714793474

[1714793474]

1714793474

[NickW]

I think people would notice if the shares their client claimed to have submitted was different to what the pool reported.

What you suggest is not how a pool operator would steal. They would basically claim the total shares submitted was higher than the actual, meaning the payout for the real shares is less. This is why some pools show proofs of work to demonstrate this is not happening.

[SgtSpike]

I think BTCGuild was accused of this recently.

The bottom line is, yes, it is possible, and no, you couldn't detect it if it was well-hidden.

[Jack of Diamonds]

I think BTCGuild was accused of this recently.

The bottom line is, yes, it is possible, and no, you couldn't detect it if it was well-hidden.

If you use a PPS derivative reward scheme it doesn't matter.
Since the operator has only two options, to pay for your submitted shares or get labeled as a scammer

PPS fraud would be instantly visible within less than a week, since there is zero variance & multiple miners would be reporting 2-3% less than average earnings suddenly.

Maybe if you take 0.1% or a similar amount, it would be brushed off as a normal occurrence.
Doesn't sound too profitable though.

Bigger pools can already earn enough without risking their entire reputation.

[SgtSpike]

I think BTCGuild was accused of this recently.

The bottom line is, yes, it is possible, and no, you couldn't detect it if it was well-hidden.

If you use a PPS derivative reward scheme it doesn't matter.
Since the operator has only two options, to pay for your submitted shares or get labeled as a scammer

PPS fraud would be instantly visible within less than a week, since there is zero variance & multiple miners would be reporting 2-3% less than average earnings suddenly.

Maybe if you take 0.1% or a similar amount, it would be brushed off as a normal occurrence.
Doesn't sound too profitable though.

Bigger pools can already earn enough without risking their entire reputation.

Yeah, I do like the PPS model.  It's just that pools have to take so much in fees to account for the potential variance that could literally bankrupt them.

[Meni Rosenfeld]

This is easily detectable with the proper client-side mining software. Everyone just needs to make sure that whenever they find a valid block, it is reported by the pool.

The problem is, of course, that people don't use the proper software (maybe it doesn't yet exist), or care much about trying to detect this.

If you use a PPS derivative reward scheme it doesn't matter.

If you use PPS it doesn't matter.

If you use a "PPS derivative" (by which I guess you mean *MPPS) you have the same problem, operator can hide found blocks, keep their reward to himself and misrepresent the pool's balance.

Yeah, I do like the PPS model.  It's just that pools have to take so much in fees to account for the potential variance that could literally bankrupt them.

This is fixable with some protocol changes, adoption of decentralized metapools, and careful risk assessment (I'll write something about that last part pretty soon).

[Jack of Diamonds]

Yes, *MPPS or anything which bases payments on found blocks will still have the same problem.

I meant any kind of PPS variant that pays a fixed amount of BTC per share, and is withdrawable regardless of the pool finding blocks or not.
(Though so far I have only run into 50 / difficulty minus various fees)

[Sukrim]

I think BTCGuild was accused of this recently.

The bottom line is, yes, it is possible, and no, you couldn't detect it if it was well-hidden.

If you use a PPS derivative reward scheme it doesn't matter.
Since the operator has only two options, to pay for your submitted shares or get labeled as a scammer

No, he has a third option:
Relay every 1000th share from another pool to his own miners and always answer "invalid" to the solution sent from the miner. while relaying the correct solution back to the other pool/solo miner.

This would increase stale shares by only 0.1% (not really detectable) while increasing the operator's income potentially a lot (a lot of pools only take the mined transaction fees as income, which are ~0.5% of a block - if you can earn 0.6% instead, you get 20% more!).

Of course you could also tune this, depending on the true stale share rate - the lower it is in reality, the more the operator is free to cheat.

This can ONLY be detected, if miners monitor the true difficulty of the shares they submit (which none do I think, as it costs a bit of CPU power) and also pay close attention to the blocks that the pool solves.

As miners cannot currently choose or verify which transactions go in their getworks, there's no way for them to be sure if they don't find the block themselves.

[Artefact2]

Relay every 1000th share from another pool to his own miners and always answer "invalid" to the solution sent from the miner. while relaying the correct solution back to the other pool/solo miner.

Shares don't work that way. This can't be done.

[Sukrim]

I can forward getwork requests and the answer if the getwork was accepted or not is being issued by the pool, not a local instance. What would not work exactly?

[Sukrim]

Arsbitcoin uses SMPPS, not PPS!
There is a possibility that you get paid less than PPS with that scheme, so it is NOT 100% comparable to PPS.

The pool itself is nice, it's just not PPS.

@Artefact2:
I meant it like this:

external getwork source --> Pool --> Miner --"Solution"--> Pool --"Solution"--> external getwork source --"share valid"--> Pool --"share rejected"--> Miner